1. Field of the Invention
The present invention relates to an image processing method, an image processing device, an image forming device, and an inkjet recording device, and particularly relates to a quantization processing technology that converts multi-gradation image data into image data of a gradation number less than an original gradation number.
2. Description of the Related Art
An inkjet recording device that forms a color image on a recording medium has a configuration including an inkjet head of a full line type in which nozzles are installed over a length corresponding to an entire width of the recording medium.
In a single pass scheme, a full-line type inkjet head and a recording medium are relatively moved only once, and an image is formed over a whole surface of the recording medium. According to this single pass scheme, it is possible to perform image formation at higher speed than a serial scheme in which the head is carried in a main scanning direction to perform image formation in the same direction while the recording medium is intermittently sent at predetermined feed pitches in a sub-scanning direction.
In inkjet image formation of the single pass scheme, when an abnormal nozzle occurs like abnormality of an ejection (flying) direction, abnormality of an ejection droplet amount and non-ejection, and so on, a streak (streak-like density unevenness) so occurs. There is known a technique that masks the abnormal nozzle, applies density correction to a normal nozzle near the abnormal nozzle and lowers the visibility of the streak in order to suppress the occurrence of the streak resulting from the occurrence of the abnormal nozzle.
Japanese Patent No. 4604614 (PTL 1) discloses a technology that lowers the streak visibility by converting a defective nozzle pixel into a minimum density and assigning a density exchange table of higher density than peripheral nozzles of a defective nozzle.
Japanese Patent Application Laid-Open No. 2004-202795 (PTL 2) discloses a technology that selects a dither matrix based on the position of a defective nozzle and performs quantization processing.
Japanese Patent No. 4670696 (PTL 3) discloses an image processing method that prevents an occurrence of an artifact in an output image by acquiring non-ejection position information in quantization processing to generate pseudo-gradation data from input image data and replacing a basic threshold matrix with a sub-matrix for non-ejection (a threshold matrix that realizes dot arrangement in which a streak due to non-ejection is not remarkable) with respect to a peripheral region including the non-ejection position.
Japanese Patent No. 4614076 (PTL 4) discloses a technology that corrects a threshold matrix based on an error in nozzle ejection characteristics, reduces image density unevenness and reduces graininess in digital half-toning using the threshold matrix.